| China’s forest resources are limited, the total amount of resources is relatively insufficient, low land productivity, poor management science, technology, support and low intensity factors of the rapid development of China’s forestry. However, with the fast development of image processing and computer vision and other new technologies, and the emergence of new types of equipment, in order to improve the scientific management and efficient use of existing forest resources provide a strong technical support.From the research methods of image processing and computer vision technology-based view, By analyzing the spatial variation of four rectangular information points in 1 years, the experimental data of the traditional measurement methods are compared with the experimental data, which confirms the consistency of the two methods, and is also consistent with the annual cycle of tree growth. At the same time, by measuring the incremental changes in the height of the tree height of Grohe, and compared with the measurement results of the visual technology and traditional methods, the results show that the growth rate of the three measurement results are consistent.Therefore, using visual technology method to measure the tree information point. It can realize the monitoring study on tree height and volume increment changes through growth in the proportion of incremental changes of, for the next step to achieve on tree growth amount inversion, remote wireless monitoring and promote the rapid development of forestry information laid the good foundation of theory and technology..The main innovation of this paper and the contents are as follows:(1) For the first time set up for capturing images of trees parallel binocular vision platform. In the experimental design process, with two identical MV-VS078FC high resolution industrial CCD camera, the choice of AFT-0814MP is the camera lens, and connected with the MV-1394 image acquisition card, to complete the image acquisition target trees, and containing scale the horizontal ruler to adjust the baseline distance between the camera and at the same time with the DS3 and J2 theodolite level two measurements guarantee the same position, are secured by a rigid tripod, to complete the set up parallel binocular vision platform. Take the same position, different time points to collect the tree images and traditional measurement(2) Extracting information point coordinates with membership degree merging algorithm and 2D maximum entropy theory. In the experiment, the image of the tree is labeled with 4 rectangular red information points, and the image of the tree containing the information points is analyzed and processed by using the mean and median filter processing. At the same time, the image information points of trees, the maximum membership degree and the processing, ensure the full coverage of the information point, in the process of binarization, using one-dimensional, two-dimensional maximum entropy algorithm, to ensure the accuracy of image threshold selection.(3) Spatial distance of information points and tree year. In the process of data processing, the center of information points is used instead of the center of each image information center. The pixel coordinates of the image plane are determined, and the corresponding physical coordinates are transformed according to the parameters of the camera. At the same time, with the imaging characteristics of parallel binocular vision platform, the center coordinates of image information points and the space distance between the centers of each information point are calculated, and the measurement results are compared with the annual cycle of tree growth.(4) Study on the inversion process of tree growth. According to the incremental changes tree information horizontal distance of each measuring the properties of proportion transform combined with the changes and changes of DBH information, calculation formula and one yuan and two yuan in volume, the volume change of each measuring time, and the change of growth of tree height, so as to achieve the information through the horizontal distance increment change to achieve change tree height and volume change of each measurement time, complete inversion of the growth of trees. If the first known measurement DBH and tree height, can be obtained in each measurement, the specific increase in tree height and volume weight. It is proved that the method and the method of image processing based on two dimensional entropy algorithm and binocular vision technology are accurate, and the test design is feasible, and the measurement results are credible.(5) Comparison of results of different measurement methods. According to the interval of 1 years,25 times the tree image information collected by the spatial distance, and high position information chest horizontal distance increment, calculate the change of the change of tree height and stem volume of each measurement; and compared with the traditional measurement of DBH incremental changes in contrast induced volume change and altimeter measured the seasonal change of tree height data, confirmed the different measuring methods for data consistency, it is in line with the annual variation of trees. Therefore, according to the information at various levels in the process of measuring the change of distance, can realize inversion changes of tree height and volume growth of trees at each measurement.(6) Validation studies of other tree species. Single species data may be very limited, lack of convincing, in order to fully demonstrate the feasibility of vision measurement method, at the same time in the measurement of each collection, traditional measurement and altimeter of quarter tree image selected Liu Shuhe Yushu two species as a reference, highly seasonal measurement. From the two point range tree information and results of altimeter measurements of the data, changes with its growth rate is similar, that take two-dimensional entropy method and algorithm of binocular vision technology combined with other species in the measurement is also feasible.In short, the membership and the merging algorithm of 2-D maximum entropy and the theory of binocular vision technology, can effectively extract the spatial information and image information of trees, tree height and volume changes in the data obtained by the space level information distance increment, practice proves that the results are accurate and reliable, is consistent with the the annual tree growth cycle, and the traditional method of measurement result is consistent, is also consistent with other species of measurement results. This shows that, by using the two-dimensional entropy algorithm and binocular vision technology, it can be based on the distance between the information points to achieve the growth of tree.At the same time, through the study of tree image growth state, we can know the process of tree growth,provide theoretical and practical support for the development of image feature extraction and research. It can provide some data support and technical support for improving tree species and planting adaptation to local environment.
|
PDF Full Text Request